ORIGINAL ARTICLE. Thierry Bové a, *, Yves Van Belleghem a, Katrien Franc ois a,frankcaes a,micheldepauw a, Yves Taeymans a and Guido J.

Transcription

1 Interactive CardioVascular and Thoracic Surgery 24 (2017) doi: /icvts/ivx028 Advance Access publication 2 March 2017 ORIGINAL ARTICLE Cite this article as: Bové T, Van Belleghem Y, Franc ois K, Caes F, De Pauw M, Taeymans Y et al. Low target-inr anticoagulation is safe in selected aortic valve patients with the Medtronic Open Pivot mechanical prosthesis: long-term results of a propensity-matched comparison with standard anticoagulation. Interact CardioVasc Thorac Surg 2017;24: a b Low target-inr anticoagulation is safe in selected aortic valve patients with the Medtronic Open Pivot mechanical prosthesis: long-term results of a propensity-matched comparison with standard anticoagulation Thierry Bové a, *, Yves Van Belleghem a, Katrien Franc ois a,frankcaes a,micheldepauw a, Yves Taeymans a and Guido J. Van Nooten b Heart Center, University Hospital Ghent, Ghent, Belgium University Ghent, Ghent, Belgium * Corresponding author. Cardiac Surgery, University Hospital Ghent, De Pintelaan 185, 9000 Ghent, Belgium. Tel: ; thierry.bove@ugent.be (T. Bove). Received 23 May 2016; received in revised form 7 December 2016; accepted 2 January 2017 Abstract OBJECTIVES: To investigate the long-term results of a low international normalized ratio (INR)-anticoagulation program in selected patients after aortic valve replacement (AVR) with the Medtronic Open Pivot mechanical heart valve (OPMHV). METHODS: From January 1993 to December 2012, 909 OPMHV valves were used for single AVR. Patients with preserved sinus rhythm and left ventricular function (Low-INR, n = 552), were managed to an INR of and compared to patients (Standard-INR, n = 357) treated with standard anticoagulation (INR ). Long-term outcome was analysed for survival and valve-related events, on propensity score matched pairs of 169 patients/group. RESULTS: Within a follow-up cumulating 3096 patient-years, 10- and 15-year survival was significantly better for Low-INR patients: 79% and 63% vs 63% and 34% (P < 0.001). Multivariate analysis of late mortality identified older age [odds ratio (OR) = 1.05], chronic pulmonary disease (OR = 1.90) and coronary artery disease (OR = 1.57) as patient-related risk factors, and erratic INR (OR = 2.57) as anticoagulation-related factor. The linearized thromboembolic rate was 0.72%/patient-year in Low-INR patients, vs 0.87%/patient-year in Standard-INR patients (P = 0.59), revealing INR variability as sole predictor (OR 3.54, 95% confidence interval (CI) , P = 0.022). The linearized bleeding incidence was respectively 0.61%/patient-year and 1.21%/patient-year for Low-INR and Standard-INR patients (P = 0.04), retaining older age (OR 1.06, 95% CI , P = 0.009), hypertension (OR 2.06, 95% CI , P = 0.05) and erratic INR (OR 9.83, 95% CI , P < 0.001) as independent risk factors. CONCLUSIONS: This study demonstrated that application of an anticoagulation regimen, aiming a low INR, individualized to selected aortic OPMHV patients, is safe and effective over more than 20 years, without increasing the thromboembolic complication rate while lowering the haemorrhagic events. However, INR variability remains worrisome because of its deleterious effect on outcome. Keywords: Mechanical aortic valve prosthesis Outcome analysis Anticoagulation INTRODUCTION The use of mechanical valve prostheses is inevitably linked to the lifelong requirement of anticoagulant medication. Hereto, only vitamin K-antagonists have proven to be effective, yielding however, the major drawback of a narrow therapeutic range, that necessitates frequent monitoring and dose adjustments to limit the clinical consequences of the pendulous risk between thrombotic and haemorraghic complications. Since it has been shown that the hinge mechanism in bileaflet mechanical valves is critical in the initiation of platelet activation and thrombus formation [1, 2], the design of the leaflet housing has become an essential inter-player in the anticoagulation management. The housing mechanism of the Open Pivot mechanical heart valve (OPMHV; Medtronic, Minneapolis, MN, USA) is specifically constructed to minimize cavitational shear stress areas. This prosthesis was introduced at our institution when it first became commercially available in 1992 [3]. Based on the favourable effects on thrombogenicity after in vitro testing [4, 5], we actively introduced, as from 1993, a change in anticoagulation regimen by arbitrarily dividing candidates for aortic valve replacement (AVR) with a mechanical prosthesis into 2 categories, depending on their presumed risk for valve-related adverse VC The Author Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.

2 T. Bové et al. / Interactive CardioVascular and Thoracic Surgery 863 events. Patients with preserved left ventricular (LV) function in sinus rhythm, and without major associated vascular disease were considered to be at low risk, and the anticoagulation was prospectively lowered to maintain the international normalized ratio (INR) between 1.5 and 2.5. In other patients, the INR was routinely kept between 2.5 and 3.5 as a standard regimen. Respecting meanwhile a more than 20-year experience with the use of this mechanical prosthesis [6], this study is aiming to assess whether such policy targeting a lower anticoagulation profile in patients requiring a single AVR is sustainable in terms of safety and efficacy over a long-term period. MATERIAL AND METHODS Study population Between January 1993 and December 2012, 909 consecutive patients underwent AVR with the OPMHV. Patients were appointed to a different anticoagulation protocol, mainly based on the arbitrary use of 2 criteria, i.e. preservation of sinus rhythm and systolic LV function. Five hundred fifty-two patients were treated to a target INR of (Low-INR group), while 357 patients received the standard regimen to obtain an INR of (Standard group). Our standard operative technique for AVR with a mechanical prosthesis has been described previously [3], with all valves implanted routinely in supra-annular position with single interrupted stitches, without any use of pledgets. All patients were entered prospectively in a homemade database. The study was approved by the institution s Ethical Committee (EC/2015/0288) waiving the need for informed consent. LV function was determined by echocardiography, retaining a left ventricular end-diastolic diameter >60 mm to categorize patients into the standard anticoagulation program, instead of using LV ejection fraction, which was not always available in this retrospective analysis. Atherosclerosis was defined by prior peripheral vascular surgery or documented vascular disease, or evidence of calcific aortic disease at intraoperative echocardiography. The study follow-up was ended on 30 June 2015, reaching 99% completeness. Follow-up time averaged 10.2 ± 5.1 years, ranging from 1.8 to 23.1 years for the surviving patients, and yielded in total patient-years. Only 6 patients residing in foreign countries were lost. In total, 301 (65%) of all hospital survivors consulted the authors institution on a yearly basis. For the remaining, the clinical and laboratory data were gathered from general practitioners or specialists. Twenty-eight Low-INR patients were excluded from further analysis, after an additional valve procedure (n = 15), while the original valve was replaced in 11, and 2 patients underwent a heart transplantation. Another 38 patients developed persistent atrial fibrillation during follow-up, in average 9 years after the initial surgery. Consequently anticoagulation was re-adapted to a higher regime and henceforth follow-up in the low-inr group was stopped. Eight patients ceased anticoagulation because of repetitive haemorrhage (n = 2), while 3 patients were shifted to anti-platelet therapy due to advanced age, lack of compliance or psychological disorders, while in some cases the decision was taken without evident reason and patients were urged to restart their anticoagulation. Hundred and thirty patients had additional anti-platelet treatment. A total of INR-values were gathered (minimal 50 values per patient), and analysed longitudinally. INR variability was measured for the whole population and represented by the mean value and the standard deviation. This was repeated for all INR-values of each individual patient. The INR was considered erratic for a patient if the mean INR-value exceeded at least 1 standard deviation of the mean INR-value of the whole group. Statistical analysis Statistics were performed with SPSS 24.0 (IBM, Chicago, Ill, USA). To adjust for basic demographic differences, propensity score matching was done for the 552 Low-INR and 357 Standard-INR patients, based on a logistic regression model with INR-treatment as dependent variable, and age, gender, preoperative New York Heart Association (NYHA) class, renal dysfunction, chronic obstructive lung disease, preoperative sinus rhythm versus atrial fibrillation, LV function, coronary artery disease or other associated atherosclerosis, and need for emergency surgery as independent variables. This resulted in 169 matched pairs for each group, obtaining a 1 1 matching without replacement, and using a standard difference (caliper) of 0.1 between groups. The fit of the propensity score matching was graphically assessed by the distribution of the propensity scores, and by indicating the standardized difference of the included variables before and after matching (Fig. 1A and B). Subsequent outcome analysis was performed on the matched cohorts, yielding a follow-up of 3096 patient-years (mean 9.2 ± 5.4 years, maximum 21.3 years). Continuous data are expressed as mean and standard deviation. Categorical variables are shown as number and frequency. Comparison of continuous parameters was based on the unpaired Student s t-test for parametric data, and the Wilcoxontest for non-parametric data. Categorical data were compared via Chi-square or Fisher s exact test. Post hoc tests were performed when necessary, applying the Bonferroni correction. The definition of complications was according to the published guidelines by Akins and co-workers for reporting valve-related morbidity and mortality [7]. Early outcome was defined as inhospital mortality. Late outcome focused on survival and freedom from valve- and anticoagulation-related events. Variables entered into the risk analysis for survival included: age as continuous variable; age >65 years; gender; emergency surgery; associated coronary artery bypass grafting; redo-surgery; hypertension; diabetes; LV dysfunction by left ventricular end-diastolic diameter >60 mm; NYHA class; chronic renal failure defined as glomerular filtration <60 ml/ min/body surface area; chronic obstructive lung disease; atherosclerosis; endocarditis; thrombosis; thromboembolism; bleeding and erratic INR. The survival or event-free survival was analysed using the Kaplan Meier product-limit estimation. Univariate comparison of survival and freedom from valve-related events between the matched study groups was performed with the log-rank test (Mantel-Cox). Subsequently, risk factor assessment of the endpoints survival, freedom from thromboembolism and from bleeding was done by multivariate Cox-regression analysis (backward stepwise method) for the entire matched cohort, irrespective of the grouping variable. The level of significance was set at P <0.05 RESULTS Patient- and operation-related results While the characteristics of the unmatched groups are clearly different, the preoperative demographics are globally comparable ADULT CARDIAC

3 864 T. Bové et al. / Interactive CardioVascular and Thoracic Surgery Figure 1: (A) Distribution of the propensity scores between the groups (group 0 = Standard-INR and group 1 = Low-INR) before and after propensity score matching. (B) Standardized differences of the variables included in the propensity score model, before and after matching. for the matched pairs (Table 1). According to the treatment intention, the mean INR was 1.90 ± 0.72 in the Low-INR group vs 2.86 ± 1.90 in the Standard-INR group (P < 0.001), with a rate of erratic INR of 3.6% and 9.5%, respectively (P = 0.05). The size of all implanted valves were equally distributed over both groups (P = 0.88) (Fig. 2). The OPMHV valve was used for AVR in 11% redo-operations (Low-INR = 6% and Standard- INR = 17%, P = 0.004), and 5% Bentall procedures (Low-INR = 2% and Standard-INR = 6%, P = 0.12). Associated procedures were 47% coronary artery bypass grafting (Low-INR = 42% and Standard-INR = 50%, P = 0.10) and 5% mitral valve repair (Low- INR = 3% and Standard-INR = 7%, P = 0.05). Survival In-hospital mortality was 0.3% (n = 1), and overall survival at 90 days was 97% (CI %). Survival analysis showed a significantly better survival of 98 ± 2%, 92 ± 4%, 79 ± 6% and 63 ± 10% in the Low-INR group, in comparison to 91 ± 4%, 79 ± 6%, 63 ± 8% and 34 ± 10% survival in Standard-INR patients at 1, 5, 10 and 15 years, respectively (P < 0.001) (Fig. 3). Univariate analysis revealed age (P = 0.003), coronary artery disease (P = 0.001), chronic obstructive lung disease (P = 0.036), advanced NYHA class at the time of surgery (P < 0.001), and erratic INR (P = 0.028) as determinants of survival. By multivariate analysis, independent risk factors for late mortality were older age (OR 1.05, 95% CI , P < 0.001), coronary artery disease (OR 1.57, 95% CI , P = 0.019), erratic INR (OR 2.57, 95% CI , P = 0.002) and chronic obstructive lung disease (OR 1.90, 95% CI , P = 0.021). Valve thrombosis During follow-up, 5 events of valve thrombosis were recorded, 2 in the Low-INR group and 3 in the Standard-INR group (P = 0.65). Four patients were treated successfully with thrombolytics, while 1 haemodynamically unstable patient whose thrombosed aortic prosthesis was additionally affected by endocarditis, underwent urgent surgery but died. This event occurred in average 69 ± 52 months after initial implantation, with a subtherapeutic INR in all cases (mean INR = 1.13; range ). The linearized rate of valve thrombosis was calculated at 0.11% per patient-years in Low-INR patients, compared to 0.24% per patient-years in Standard-INR patients. Thromboembolism Thromboembolic events were noticed 12 times in the Low-INR group, compared to 11 episodes in the Standard-INR group, and included persistent cerebrovascular accidents (Low-INR 8, Standard-INR 5), transient ischaemic attack (Low-INR 3, Standard-INR 5), and peripheral arterial embolism (Low-INR 1, Standard-INR 1). The majority of events occurred within the first 5 years (33 ± 18 months) after valve implantation, with an average INR at the event time of 1.6 ± 0.1. The linearized rate of thromboembolism was 0.72% per patient-year in Low-INR patients vs 0.87% per patient-year in Standard-INR patients. Freedom from thromboembolism was similar for both groups (P = 0.59) (Fig. 4). Multivariate analysis maintained only erratic INR (OR 3.54, 95% CI , P = 0.022) as independent risk factor.

4 T. Bové et al. / Interactive CardioVascular and Thoracic Surgery 865 Table 1: Patient demographics Unmatched groups Matched groups Low-INR (%) Standard (%) P-value Low-INR (%) Standard (%) P-value Number Mean age (years) 58 ± ± ± 9 65 ± Gender (n, %) Male 399 (74) 264 (74) 123 (73) 121 (72) Female 153 (26) 93 (26) 46 (27) 48 (28) Etiology (n, %) Degenerative 457 (81) 288 (81) 137 (81) 135 (80) Rheumatic 9 (2) 14 (5) 5 (3) 8 (5) Endocarditis 51 (9) 22 (7) 15 (9) 15 (9) Congenital (bicuspid) 21 (4) 11 (4) 7 (4) 7 (4) Miscellaneous 13 (4) 9 (3) 5 (4) 4 (2) NYHA class (n, %) 0.67 I 6 (1) 0 (0) 3 (2) 1 (1) II 331 (60) 121 (34) < a 85 (50) 77 (45) III 182 (33) 151 (43) 71 (42) 78 (46) IV 33 (6) 85 (23) < a 10 (6) 13 (8) Rhythm (n, %) < Sinus 552 (100) 233 (65) 169 (100) 160 (85) Non-sinus 0 (0) 124 (35) 0 (0) 9 (13) Coronary artery disease (n, %) 118 (21) 161 (45) (44) 90 (53) 0.11 Hypertension (n, %) 128 (23) 86 (23) (23) 44 (26) 0.45 Diabetes (n, %) 38 (7) 91 (25) (5) 15 (9) 0.13 Left ventricular dilatation (n, %) 49 (8) 81 (23) (2) 9 (5) 0.07 COLD (n, %) 59 (10) 41 (11) (9) 18 (11) 0.47 CRF (n, %) 28 (5) 16 (5) (4) 11 (7) 0.21 Arteriosclerosis (n, %) 157 (26) 189 (53) (40) 77 (48) 0.32 Emergency (n, %) 38 (7) 44 (12) (6) 14 (9) 0.40 ADULT CARDIAC COLD: chronic obstructive lung disease; CRF: chronic renal failure; NYHA: New York Heart Association. a Bonferroni correction. Figure 2: Size distribution of the mechanical valves (mm diameter) in both groups. Haemorrhage Fourteen major bleeding events occurred during follow-up in Low-INR patients, compared to 26 in the Standard-INR group, and Figure 3: Survival plot of the Low-INR and Standard-INR patients, showing a survival benefit for Low-INR patients (logrank P < 0.001). consisted of cerebral (Low-INR 6, Standard-INR 12), gastro-intestinal (Low-INR 4, Standard-INR 7), peripheral (Low-INR 2, Standard- INR 4), traumatic (Low-INR 1, Standard-INR 3) and intraabdominal (Low-INR 1) haemorrhages. These occurred at random throughout the follow-up, at an average INR of 3.6 ± 1.0 at the

5 866 T. Bové et al. / Interactive CardioVascular and Thoracic Surgery Figure 4: 20-year freedom from thromboembolism in Low-INR and Standard- INR patients (logrank P = 0.59). event time. The bleeding event happened in 11 patients who were additionally treated with anti-platelet drugs. The linearized incidence of bleeding was 0.61% per patient-year in Low-INR patients, and 1.21% per patient-year in Standard-INR patients. Freedom from bleeding was significantly better in Low-INR patients (P = 0.04) (Fig. 5). Univariate predictors of haemorrhage were older age (P = 0.043), hypertension (P = 0.050), chronic renal failure (P = 0.002) and erratic INR (P = 0.012). Multivariate analysis only maintained older age (OR 1.06, 95% CI , P = 0.009), arterial hypertension (OR 2.06, 95% CI , P = 0.05) and erratic INR (OR 9.83, 95% CI , P < 0.001) as independent risk factors. Endocarditis Based on clinical and echocardiographic criteria, prosthetic valve endocarditis was observed in 7 patients, of whom 2 succumbed. Five underwent successful redo-avr, in average 61 ± 42 months after the first surgery. The linearized incidence of endocarditis was 0.09% per patient-year for the whole population. Paravalvular leakage and haemolysis Seven patients presented with paravalvular leakage and/or haemolysis (n = 2) in absence of any infective episode. In 5 patients, the leakage was successfully closed by surgery, without the need for valve replacement, while the latter was performed in 2 patients. Valve structural failure To date, no structural valve deterioration has occurred in the entire series. Figure 5: 20-year freedom from bleeding in Low-INR and Standard-INR patients (logrank P = 0.04). DISCUSSION The application of a low-intensity anticoagulation protocol for mechanical heart valves has been studied by several groups, including ourselves [8 12]. This study is the first to report on the results of an anticoagulation policy targeting a lower INR over a longer time span up to 20 years in selected AVR patients with the unique use of the OPMHV prosthesis. The results revealed a linearized rate of 0.72% per patient-year for thromboembolic events, and an annual risk for bleeding of 0.61% per patient, reassuring the long-term efficacy and safety of this management. It is known that both end-points are balancing the adequacy of the anticoagulation strategy in an opposite direction, i.e. increasing the INR will decrease the risk of thrombotic events but increase the risk of haemorrhage and vice versa. Whereas bleeding is mainly dependent on patient- and anticoagulation-related factors, thromboembolism or thrombosis is additionally affected by the type of mechanical prosthesis. Previous research has highlighted the favourable effect of the OPMHV hinge system on thrombogenicity [4, 12], but concurrently low incidences of thromboembolic events have also been reported with other bileaflet prostheses, although respecting much shorter follow-up times. Palatianos et al. and Chambers et al. [13, 14] published respectively a thromboembolic event rate of 0.88% per patientyear at 5 years, and 0.6% per patient-year at 12 years, with the On-X valve. In contrast, Emery et al. [15] reported a higher incidence of 1.9% per patient-year for AVR with the St Jude Medical prosthesis, over 25 years but in unselected patients. Regarding the incidence of haemorrhagic events, it is evident that a lower target-inr driven program leads to a significant decrease of bleeding problems. Although others reported a comparable rate of major bleeding events of % per patientyear [14, 16], the recent PROACT trial with the On-X mechanical valve, pointed on a major bleeding rate of 1.48% per patient-year in the group receiving a lower dose warfarin, and even 3.31% per patient-year in the control group. This complication rate seems unexpectedly elevated, especially when the use of home INRmonitoring and the high medical compliance of the enrolled

6 T. Bové et al. / Interactive CardioVascular and Thoracic Surgery 867 patients is forwarded as a solid argument to support this reduced anticoagulation treatment [11]. Independent risk factors for bleeding in our series were older age and hypertension. In patients aging over 65 years at the time of AVR, the risk of bleeding increased with 60% once the anticoagulation was outside the target zone. Considering that the age of aortic valve patients will undeniably rise with time, eventually concomitant to an increasing frequency of hypertension, care should be taken not to overtreat these elderly, by down-regulating the anticoagulant dosage [17]. The addition of anti-platelet drugs has been proposed as a useful adjunct, especially in patients with documented atherosclerotic disease, to pursue a beneficial effect on the rate of thromboembolism, while lowering the INR [18, 19]. However, according to Altman et al., we also observed some major to even fatal haemorrhagic complications when anti-platelet therapy was added; therefore, its usefulness needs to be weighed carefully on an individual basis. The time-related risk of bleeding was significantly lower in patients treated at the low target INR, partially due to a smaller range of INR variability. Increased INR variability, defined by erratic INR in our series, appeared to be the strongest predictor of developing either a bleeding or thromboembolic complication. This underscores that, regardless of the used anticoagulation policy, close surveillance of the INR is still mandatory, due to the intrinsic thrombogenicity of any mechanical valve on one hand, as well as to the detrimental effect of outranged INR variability on bleeding events, as likely reported in other non-valvular diseases requiring anticoagulation therapy [16, 20]. The strength of the initial patient selection to be included in a low INR-anticoagulation program at the time of AVR with a mechanical prosthesis, is of utmost importance. The differentiation to address patients in a low INR regimen was merely based on the presence of preserved sinus rhythm and LV function. Patients estimated of having an increased risk profile, were inherently treated with the standard anticoagulation protocol. Even though propensity score matching allowed to adjust for major differences in baseline patient characteristics, inherently induced by the intention-to-treat principle, the late survival was still in favour of AVR patients treated to a low INR. Independent of the anticoagulation regimen, late survival is compromised by comorbidities such as chronic renal and respiratory insufficiency, associated coronary disease and older age. Hence, despite the statistical attempt to minimize the impact of selection bias, patients included in the standard INR regimen were more likely to undergo associated procedures such as mitral valve repair and coronary artery bypass grafting, probably suggesting more advanced cardiac disease. Interestingly, erratic INR additionally affected the long-term survival, at least by doubling the risk of death (HR 2.6). Butchart et al. [21] found identically that a high INR variability decreased the survival by 30% in patients having a mechanical valve in aortic or mitral position. They noticed that patients without obvious risk profile had a survival equal to their age- and sex-matched peers. Regardless whether a low or standard anticoagulation regimen was followed, erratic INR increased the linearized risk of valve-related events, combining bleeding and thromboembolism, which both tend to occur during episodes of out-ranged INR-values. Despite the often disabling clinical effect of both complications, these were not restrained separately as independent determinants of late outcome. It is however, conceivable that the observation of a few repeated events in a same patient has interfered by affecting the constancy of the hazard function. Another explanation might be that over 20 years, patients evolve through developing additional diseases, potentially affecting survival and valve-related outcome. In this study, we tried to cope with such evolution by excluding patients with new onset atrial fibrillation from further analysis, instead of crossing them to the high-risk group. However, this was not done for other, more subtle changes of the original patient characteristics, when these did not interact with the chosen anticoagulation regimen. Skeptics on the use of mechanical valves will argue in favour of bioprostheses, because of the negative impact of a poorly controlled anticoagulation on survival. In contrast, considering the substantial rate of structural valve degeneration in bioprostheses in the younger-aged cohorts, greater efforts to manage the lifelong anticoagulation should reinforce the survival advantage with mechanical valves as shown by the only randomized trial available on a 15-year comparison between mechanical and biological valves [22]. Our data confirm that the adoption of an anticoagulation protocol, individualized to the patient, is helpful to improve the overall outcome, at least for single aortic valve replacement. Hereto, as soon as the medical decision for AVR is taken, patients candidate for a mechanical valve have to be empowered of the consequences of this medication intake. Besides insisting on compliance, the interaction with other drugs must also be emphasized, in particular in the aging population subjected to a greater chance of having associated diseases that require medication. Self-monitoring of INR has been advanced as a beneficial tool to decrease INR variability [23 25]. This application necessitates profound patient education and supplementary financial resources to acquire the still expensive measurement device. Hence, Matchar et al. [26] demonstrated in a randomized study that self-testing of INR weekly was not superior to monthly INR-testing, when performed in dedicated, high quality thrombosis clinics. Study limitations Although the low-intensity anticoagulation regimen was started on a prospective basis, the study carries the limitations of its retrospective character. It is possible that changes in patient characteristics were incompletely accounted for analysis of study outcomes, considering the more than 20 years follow-up duration. Although we feel confident with the follow-up completeness, yielding an annual in-house follow-up of at least 65%, possible under-reporting of transient adverse events may have occurred in the group followed externally. This also includes a potential bias in the gathering of INR-values, by perhaps collecting more INR-values within the target zone among motivated patients, and vice versa. However, the sufficiently large number of patients and of INR-values should strengthen the statistical findings and clinical conclusions. We are aware that investigating the hypothesis on safety and efficacy of a low target-inr protocol in mechanical aortic prosthesis is best performed in a randomized design, but the application of strict inclusion and exclusion criteria will need a huge sample size. In this study, propensity-matching has enhanced the comparability of the study groups, but equally reduced the number of matched pairs significantly. However, the clinical reality is that the intention-to-treat appointment undeniably induces differentiation between the patient groups, still entailing the possibility of incomplete elimination of confounding. This study has certainly not the pretension to impose firm recommendations, ADULT CARDIAC

7 868 T. Bové et al. / Interactive CardioVascular and Thoracic Surgery but is likely to give a reflection of a central policy on anticoagulation management in AVR patients with the specific use of this type of prosthesis. Traditionally, a surgeon is seeking to offer the best technical solution for the patient needing a mechanical valve, but for sure, to optimize his/her long-term outcome, it might be a greater responsibility to ensure that the patient receives the best possible anticoagulation management. CONCLUSION Over 20-year experience with the OPMHV mechanical valve for AVR has demonstrated excellent clinical results with few valverelated events. In patients basically selected on preservation of sinus rhythm and LV function, the long-term maintenance of a low target-inr has proven to be safe and effective, by a superiority in lowering haemorrhagic events and without increasing the rate of thrombotic complications. However, independent of the chosen anticoagulation protocol, INR variability remains worrisome because of its deleterious effect on survival and valverelated outcome. This study underscores that the anticoagulant management in AVR patients treated with this mechanical prosthesis, is best served by an individual patient-oriented approach, to offer them an optimal long-term outcome with minimal anticoagulation- and valve-related events. Funding The University Hospital of Ghent received a Euro research grant from Medtronic Inc. for the creation and use of the homemade databank. Conflict of interest: none declared. REFERENCES [1] Bluestein D, Rambod E, Gharib M. Vortex shedding as a mechanism for free emboli formation in mechanical heart valves. J Biomech Eng 2000;122: [2] Min Yun B, Aidun CK, Yoganathan AP. Blood damage through a bileaflet mechanical heart valve: a quantitative computational study using a multiscale suspension flow solver. J Biomech Eng 2014;136: [3] Van Nooten G, Caes F, Francois K, Missault L, Van Belleghem Y. Clinical experience with the first 100 ATS heart valve implants. Cardiovasc Surg 1996;4: [4] Verdonck PR, Van Nooten GJ, Van Belleghem Y. Pulse duplicator hydrodynamics of four different bileaflet valves in the mitral position. Cardiovasc Surg 1997;5: [5] Kelly SG, Verdonck PR, Vierendeels JA, Riemslagh K, Dick E, Van Nooten GG. A three-dimensional analysis of flow in the pivot regions of an ATS bileaflet valve. Int J Artific Org 1999;22: [6] Van Nooten GJ, Bove T, Van Belleghem Y, Francois K, Caes F, Vandenplas G et al. Twenty-year single-center experience with the medtronic open pivot mechanical heart valve. Ann Thorac Surg 2014;97: [7] Akins CW, Miller DC, Turina MI, Kouchoukos NT, Blackstone EH, Grunkemeier GL et al. Guidelines for reporting mortality and morbidity after cardiac valve interventions. J Thorac Cardiovasc Surg 2008;135: [8] Acar J, Iung B, Boissel JP, Samama MM, Michel PL, Teppe JP et al. AREVA: multicenter randomized comparison of low-dose versus standard-dose anticoagulation in patients with mechanical prosthetic heart valves. Circulation 1996;94: [9] Koertke H, Zittermann A, Tenderich G, Wagner O, El-Arousy M, Krian A et al. Low-dose oral anticoagulation in patients with mechanical heart valve prostheses: final report from the early self-management anticoagulation trial II. Eur Heart J 2007;28: [10] Meschengieser SS, Fondevila CG, Frontroth J, Santarelli MT, Lazzari MA. Lowintensity oral anticoagulation plus low-dose aspirin versus high-intensity oral anticoagulation alone: a randomized trial in patients with mechanical prosthetic heart valves. J Thorac Cardiovasc Surg 1997;113: [11] Puskas J, Gerdisch M, Nichols D, Quinn R, Anderson C, Rhenman B et al. Reduced anticoagulation after mechanical aortic valve replacement: interim results from the prospective randomized on-x valve anticoagulation clinical trial randomized Food and Drug Administration investigational device exemption trial. J Thorac Cardiovasc Surg 2014;147: ; discussion [12] Van Nooten GJ, Van Belleghem Y, Caes F, Francois K, Van Overbeke H, Bove T et al. Lower-intensity anticoagulation for mechanical heart valves: a new concept with the ATS bileaflet aortic valve. J Heart Valve Dis 2003;12: ; discussion 02. [13] Palatianos GM, Laczkovics AM, Simon P, Pomar JL, Birnbaum DE, Greve HH et al. Multicentered European study on safety and effectiveness of the On-X prosthetic heart valve: intermediate follow-up. Ann Thorac Surg 2007;83:40 6. [14] Chambers JB, Pomar JL, Mestres CA, Palatianos GM. Clinical event rates with the On-X bileaflet mechanical heart valve: a multicenter experience with follow-up to 12 years. J Thorac Cardiovasc Surg 2013;145: [15] Emery RW, Krogh CC, Arom KV, Emery AM, Benyo-Albrecht K, Joyce LD et al. The St. Jude Medical cardiac valve prosthesis: a 25-year experience with single valve replacement. Ann Thorac Surg 2005;79:776 82; discussion [16] Hering D, Piper C, Bergemann R, Hillenbach C, Dahm M, Huth C et al. Thromboembolic and bleeding complications following St. Jude Medical valve replacement: results of the German Experience With Low-Intensity Anticoagulation Study. Chest 2005;127:53 9. [17] Arom KV, Emery RW, Nicoloff DM, Petersen RJ. Anticoagulant related complications in elderly patients with St. Jude mechanical valve prostheses. J Heart Valve Dis 1996;5: [18] Cappelleri JC, Fiore LD, Brophy MT, Deykin D, Lau J. Efficacy and safety of combined anticoagulant and antiplatelet therapy versus anticoagulant monotherapy after mechanical heart-valve replacement: a metaanalysis. Am Heart J 1995;130: [19] Turpie AG, Gent M, Laupacis A, Latour Y, Gunstensen J, Basile F et al. A comparison of aspirin with placebo in patients treated with warfarin after heart-valve replacement. New Engl J Med 1993;329: [20] Sanden P, Renlund H, Svensson PJ, Sjalander A. Bleeding complications and mortality in warfarin-treated VTE patients, dependence of INR variability and ittr. Thromb Haemostasis 2017;117: [21] Butchart EG, Payne N, Li HH, Buchan K, Mandana K, Grunkemeier GL. Better anticoagulation control improves survival after valve replacement. J Thorac Cardiovasc Surg 2002;123: [22] Hammermeister K, Sethi GK, Henderson WG, Grover FL, Oprian C, Rahimtoola SH. Outcomes 15 years after valve replacement with a mechanical versus a bioprosthetic valve: final report of the Veterans Affairs randomized trial. J Am Coll Cardiol 2000;36: [23] Koertke H, Zittermann A, Wagner O, Koerfer R. Self-management of oral anticoagulation therapy improves long-term survival in patients with mechanical heart valve replacement. Ann Thorac Surg 2007;83:24 9. [24] Samsa GP, Matchar DB. Relationship between test frequency and outcomes of anticoagulation: a literature review and commentary with implications for the design of randomized trials of patient self-management. J Thromb Thrombolysis 2000;9: [25] Sawicki PT. A structured teaching and self-management program for patients receiving oral anticoagulation: a randomized controlled trial. Working Group for the Study of Patient Self-Management of Oral Anticoagulation. JAMA 1999;281: [26] Matchar DB, Jacobson A, Dolor R, Edson R, Uyeda L, Phibbs CS et al. Effect of home testing of international normalized ratio on clinical events. New Engl J Med 2010;363: